Create README.md

README.md

@@ -1,3 +1,17 @@
----
-license: mit
----
+---
+license: mit
+language:
+- en
+tags:
+- vision-language
+- reward-modeling
+- benchmark
+size_categories:
+- n<1K
+---
+
+Benchmark Data for ViLBench: A Suite for Vision-Language Process Reward Modeling
+
+[arXiv](https://arxiv.org/abs/2503.20271) | [Project Page](https://ucsc-vlaa.github.io/ViLBench/)
+
+There are 600 data collected from 5 existing vision-language tasks